Marker Detection in X-Ray Images

1. A method for detecting markers within X-ray images, comprising: receiving an image sequence comprising a plurality of X-ray image frames; applying one or more directional filters to the image sequence to provide respective strength of marker features for pixels of the image sequence; determining that each of the pixels of the image sequence is a marker candidate pixel when its respective strength of marker features exceeds a first predetermined threshold value; determining which of the marker candidate pixels occur at substantially the same location within the greatest number of X-ray image frames of the image sequence; combining pixels that have been determined to be marker candidate pixels into one or more pixel clusters; computing distance between each pixel cluster with respect to each of the other pixel clusters; determining an average distance from the computed distances; defining a first marker location at a location of a pixel cluster from one or more of the clusters of contiguous pixels that includes the marker candidate pixels that have been determined to occur at substantially the same location within the greatest number of X-ray image frames; and defining a second marker location at a location of a pixel cluster from one or more of the clusters of contiguous pixels that is spaced from the first marker location by substantially the average distance.

2. The method of claim 1 , wherein the determined average distance between pixel clusters serves as an estimated distance between markers within the image sequence.

3. The method of claim 1 , wherein after combining the marker candidate pixels into one or more pixel clusters, clusters that are larger than a maximum cluster size or clusters that are smaller than a minimum cluster size are rejected as false candidates and are not considered as clusters in subsequent steps.

4. The method of claim 1 , wherein a region of interest is defined within the received image sequence and marker detection is limited to the defined region of interest.

5. The method of claim 4 , wherein defining the region of interest includes: summing pixel intensity values along columns of the image sequence; applying a band pass filter to detect discontinuity among the columns of summed pixel intensities; defining vertical bounds for the region of interest as the columns of detected discontinuity; summing pixel intensity values along rows of the image sequence; applying a band pass filter to detect discontinuity among the rows of summed pixel intensities; and defining horizontal bounds for the region of interest as the rows of detected discontinuity.

6. The method of claim 1 , wherein the one or more directional filters are each applied in a corresponding direction.

7. The method of claim 1 , wherein determining which of the marker candidate pixels occur at substantially the same location within the greatest number of X-ray image frames of the image sequence includes establishing a confidence function by determining how many frames a given marker candidate pixel occurs within a fixed neighborhood.

8. The method of claim 1 , wherein in computing the distance between pixel clusters, distance is measured between respective centers of mass.

9. The method of claim 1 , wherein determining an average distance from the computed distances includes finding the mode average of the distances.

10. The method of claim 9 , wherein computing the distance between each pixel cluster with respect to each of the other pixel clusters includes generating a distance histogram representing the distances between the pixel clusters and finding the mode average of the distances includes finding the distance associated with the peak of the histogram.

11. The method of claim 1 , wherein markers are detected at the first and second marker locations.

12. The method of claim 1 , additionally comprising: determining that each of the pixels of the image sequence within a predetermined neighborhood of the first marker location is a second stage marker candidate pixel when its respective strength of marker features exceeds a second predetermined threshold value that is less than the first predetermined threshold value; combining the second stage marker candidate pixels into one or more second stage pixel clusters; determining a marker candidate score for each possible pairing of second stage pixel clusters, wherein the marker candidate score is indicative of a degree to which the pair of second stage pixel clusters resembles actual markers; and detecting markers at the locations of the pair of second stage pixel clusters that has the highest marker candidate score.

13. The method of claim 12 , wherein the detected marker locations are rejected when the highest marker candidate score is below a predetermined minimum acceptable score.

14. The method of claim 12 , wherein the marker candidate score is calculated based on distances between second stage pixel clusters.

15. The method of claim 12 , wherein the maker candidate score is calculated based on relative locations of the second stage pixel clusters within consecutive frames.

16. The method of claim 12 , wherein the maker candidate score is calculated based on a degree to which the shape of the second stage pixel clusters resembles a profile template representing an expected marker shape.

17. The method of claim 12 , wherein the maker candidate score is calculated based on a component size of the second stage pixel clusters.

18. The method of claim 12 , wherein the marker candidate score is calculated based on the output of the one or more directional filters.

19. A method for detecting markers within X-ray images, comprising: receiving an image sequence comprising a plurality of X-ray image frames; applying one or more directional filters to the image sequence; identifying a plurality of marker candidates based on the results of the directional filters; computing distance between all possible pairs of marker candidates; determining an average distance from the computed distances; detecting a first marker at a location of a marker candidate of the plurality of marker candidates that occurs at substantially the same location within the greatest number of X-ray image frames of the image sequence; and detecting a second marker at a location of a marker candidate of the plurality of marker candidates that is spaced from the first marker location by substantially the average distance.

20. The method of claim 19 , wherein identifying the plurality of marker candidates based on the results of the directional filters includes: determining that each of the pixels of the image sequence is a marker candidate pixel when its respective strength of marker features exceeds a first predetermined threshold value; and combining pixels that have been determined to be marker candidate pixels into one or more pixel clusters.

21. The method of claim 19 , wherein computing the distance between all possible pairs of marker candidates is performed with respect to the centers of mass of each marker candidate.

22. The method of claim 16 , wherein determining an average distance from the computed distances includes finding the mode average of the distances.

23. A computer system comprising: a processor; and a non-transitory, tangible, program storage medium, readable by the computer system, embodying a program of instructions executable by the processor to perform method steps for detecting markers within X-ray images, the method comprising: receiving an image sequence comprising a plurality of X-ray image frames; applying one or more directional filters to the image sequence; identifying a plurality of marker candidates based on the results of the directional filters; computing distance between all possible pairs of marker candidates; determining an average distance from the computed distances; detecting a first marker at a location of a marker candidate of the plurality of marker candidates that occurs at substantially the same location within the greatest number of X-ray image frames of the image sequence; and detecting a second marker at a location of a marker candidate of the plurality of marker candidates that is spaced from the first marker location by substantially the average distance.